Saturday, August 19th, 2023
University of California, Santa Barbara

Affiliated Event:

Crypto meets AI

ABOUT

The Glowing Hot Topics in Cryptography (GHTC) workshop series is a one-day round table event that brings together experts from both industry and academia to discuss the latest and most important developments in the field of cryptography. The round tables will feature world renowned panelists from both industry and academy. GHTC aims to provide a forum for exchanging ideas, research, and insights on hottest topics in cryptography.

The second edition of GHTC, a workshop series that focuses on cutting-edge topics in cryptography, will feature the theme "Crypto meets AI". This topic aims to explore the intersection of cryptography and artificial intelligence and examine how the two fields can work together to enhance security, privacy, and other important aspects of digital communication.

PROGRAM COMMITTEE
Paulo S. L. M. Barreto

University of Washington Tacoma, USA

View Profile
Francisco Rodríguez Henríquez
Francisco Rodríguez Henríquez

Technical Director

View Profile
Michael Scott
Michael Scott

Technical Director

View Profile
Víctor Sucasas
Víctor Sucasas

Senior Director Cryptography Engineer

View Profile
DATE AND PLACE

University of California,
Santa Barbara
View on map

Saturday
August 19th, 2023
Add to calendar
ATTENDANCE

The event will be hybrid so that attendees, panelists, and moderators can attend GHTC 2023 in person or remotely.

REGISTRATIONS

The workshop is an affiliated event of www.crypto.iacr.org/2023/. To register to the workshop, please register to CRYPTO 2023, and mark in the registration form the GHTC workshop. Please read the participation guidelines.

SCHEDULE GHTC 2023
Time Activity Place
8:00am - 12:00pm Registration Corwin Lobby
8:30am - 9:45am Breakfast De La Guerra Commons
10:00am - 10:15am Introductory Remarks Corwin West
10:15am - 10:45am Break Lagoon Plaza
10:45am - 11:45am Round Table 1:
“Crypto intersects AI”
Corwin West
11:45am - 12:45pm Round Table 2:
“Crypto meets AI: AI and cryptanalysis”
Corwin West
12:45pm - 2:05pm Lunch De La Guerra Commons
2:10pm - 3:10 pm Round Table 3:
“Privacy Preserving AI”
Corwin West
3:25pm - 3:55pm Break Lagoon Plaza
3:55pm - 4:55pm Round Table 4:
“Lessons learned”
Corwin West
5:00pm End Lagoon Plaza
PROGRAM
01/
TITLE
Crypto intersects AI
ABSTRACT
Given oracle access to a Deep Neural Network (DNN), the cryptanalytic problem of model extraction whereby the weights and architecture of a DNN are extracted or inferred using side-channel techniques and/or black box oracle accesses, is one of the hottest topics of cryptographers trying to hack AI.
CONFIRMED PANELISTS
We are happy to have the following confirmed panelists for this round table:
Shivam Bhasin

Nanyang Technological University
Singapore

sbhasin@ntu.edu.sg

Matthew Jagielski

Google (remote)

jagielski@google.com

Daniel Huynh

Mithril Security, Paris

daniel.huynh@mithrilsecurity.io

Stjepan Picek

Radboud University, The Netherlands

picek.stjepan@gmail.com

MODERATOR
Michael Scott
02/
TITLE
Crypto meets AI: AI and cryptanalysis
ABSTRACT
Machine learning techniques have been used to boost standard side-channel techniques to raise their capabilities to a much higher impact level. This line of research has seen an ever-growing number of papers accepted in applied cryptographic conferences such as CHES.
CONFIRMED PANELISTS
We are happy to have the following confirmed panelists for this round table:
Aron Gohr

New Zealand

aron.gohr@gmail.com

Stjepan Picek

Radboud University, The Netherlands

picek.stjepan@gmail.com

Fatemeh Ganji

Worcester Politechnical Institute, USA

fganji@wpi.edu

Mélissa Rossi

France (remote)

melissa.rossi@ssi.gouv.fr

Kristin E. Lauter

North American Labs, Meta AI, Seattle USA

MODERATOR
Francisco Rodriguez Henriquez
03/
TITLE
Privacy Preserving AI
ABSTRACT
After years of research on generative models and anonymization techniques, privacy for AI seems to be shifting towards novel technologies such as Fully Homomorphic Encryption (FHE) and secure Multi-Party Computation (MPC). However, both MPC and FHE bring new challenges and trade-offs. Among others, this table will discuss the following trade-offs: Will FHE semi-honest security be sufficient for AI applications, or will it require Verifiable Computation on top? Can MPC be more effective than FHE in this context, or will its interactive nature be an impairment for cloud service deployments? Also, when compared with previous approaches like generative models and anonymization techniques, are FHE and MPC costs justified? Moreover, recent research has shown that even when Privacy Enhancing Technologies (PETs) are applied to AI, specific attacks on ML algorithms are possible, e.g., training data extraction, model extraction or model poisoning. Can AI models be protected when privacy mechanisms are put in place?
CONFIRMED PANELISTS
We are happy to have the following confirmed panelists for this round table:
Ilaria Chilloti

Zama, France (Confirmed)

ilaria.chillotti@zama.ai

Daniel Huynh

Mithril Security, Paris

daniel.huynh@mithrilsecurity.io

Marcel Keller

CSIRO, Australia

Daniel Escudero

JP Morgan, USA (Confirmed)

daniel.escudero@protonmail.com

MODERATOR
Victor Sucasas
04/
TITLE
Lessons learned
ABSTRACT
In this final panel, we will review all the ideas, caveats and main comments given in the first three round tables. We will feature for this round table a selection of the panellists that participated in the previous tables.
CONFIRMED PANELISTS
We are happy to have the following confirmed panelists for this round table:
Benoit Chevallier-Mames

Zama, France

benoit.chevalliermames@zama.ai

David Gerault

CRC-TII
UAE

david.gerault@tii.ae

Daniel Escudero

JP Morgan, USA (Confirmed)

daniel.escudero@protonmail.com

MODERATOR
Francisco Rodriguez Henriquez